Design On The Software System For The Intelligent RMB Sorter

Sometimes bank needs to sort RMB according to the denomination, version time, lately old etc. Currently, most domestic banknote circulating only has simple counting and identification of authenticity functions. Therefore, the classification is still rely on manual handling work. This artificial classification not only occupy more human and classification speed, quality exist big problem.Because of higher technology content of intelligent RMB sorter, in the world only Japan, the USA, Britain, Germany and other countries can produce a few notes intelligent sorter. Most of the domestic banks imports. However, the introduction of foreign products is very expensive, the cost of maintenance services is high. And they are mostly devoted to the design of the foreign currency. So the effect is not very good for the RMB.There is therefore an urgent need for more powerful functions, the higher degree of intelligent RMB sorter. If developed suitable superior cost performance ratio intelligent RMB sorter for our country, not only have academic value, but also have practical value.This dissertation is based on detailed study the design methods of banknotes sorter in domestic and foreign. Learning advantages and avoiding shortcomings. To be directed against the drawbacks of little functions and low effect. Using fluorescence, magnetic and infrared identification of authenticity, the three theory techniques to identify. The paper makes a thorough research on the recognition method. Patter recognition, Digital image processing and computer controlling technologies are widely used in the Currency Sorter. High-speed image processing and recognition system architecture which based on the AT89C51 and TMS320VC5402 is brought out. The use of Digital image processing in this system is expatiated. Digital image denotation and pretreatment is recited, including disposing noise, emendating incline. Banknotes' histogram characteristic is analyzed. The realization tactic of median filter, emendating incline, outline pick-up arithmetic is designed. Neural network haves better sort capability, adapting to banknotes image recognition. In the recognition of the denomination, with the neural network advantage from the point of delivery as a feature vector images contours. The 50*30*5 BP neural network recognition methods is designed, the BP neural network learning rules is given. In the recognition of new and old, using notes intensity statistical value as a feature vector. The 255*8*3 LVQ neural network is designed, the LVQ neural network learning rules is given. The new and old recognition methods based on the colorized HSI formats is discussed. Finally, the overall software structure functional module program, and some other source code are debugged.